In an increasingly digitized and interconnected world, network-based modeling paradigms provide an analytical lens to understand the complex ecosystem of interactions throughout the life cycle of information flow, from how we share information to how we perform computations, collaborate, and make inference in downstream applications. My research focuses on stochastic modeling, analysis, and optimization to improve the resilience of networked systems with an emphasis on two main themes: i) enabling reliable communication and computations in distributed systems; ii) modeling, analysis, and control of contagions (such as misinformation) over social networks. The first thrust of my research focuses on improving resilience in communicating and aggregating information from distributed sources of data. The performance and scalability of emerging decentralized machine learning and communication paradigms is intimately linked to the efficient construction of sparse yet reliably connected topologies over which clients communicate and perform computations. To this end, we analyze random K-out graphs, which are receiving increasing attention for decentralized network design in diverse applications, including privacy-preserving decentralized learning algorithms and the design of next-generation internet architectures. Our recent body of work provides a formal characterization of the efficiency-connectivity trade-offs achievable with random K-out graphs. Our results highlight that random K-out graphs achieve reliable connectivity with far fewer edges compared to classical random graph models such as Erdos Renyi random graphs, even under client heterogeneity and random and targeted attacks on nodes and links. The second thrust of my research is to model, analyze, and control the widespread propagation of contagions, e.g., misinformation and pathogens, over social networks. By transforming how people communicate information, online social networks have emerged as an instrument for shaping public opinion with a direct bearing on critical issues, such as safeguarding public health and safety. Akin to different strains of a pathogen arising through mutations, information can be actively modified to increase its propensity of being shared and propagated across social media platforms. To this end, we examine mechanisms that lead to the widespread propagation of contagions and identify risk factors that can trigger such outbreaks. We discuss our recent results that investigate how the interplay of evolution and network structure ultimately impact the spreading dynamics of evolving contagions.
Mansi Sood is a Ph.D. candidate in Electrical and Computer Engineering at Carnegie Mellon University (CMU). Before this, she completed her Bachelors and Masters in Electrical Engineering at the Indian Institute of Technology (IIT) Bombay, India. Her research interests span stochastic modeling, learning, privacy, and optimization in complex socio-technical systems. A key theme in her doctoral research has been leveraging the structure of interactions to improve resilience in networked systems. She won a Best Paper Award at the IEEE International Conference on Communications (ICC) ‘21, and she has been twice recognized as an EECS Rising Star. Her work has appeared in premier venues, including the Proceedings of the National Academy of Sciences (PNAS) and IEEE Transactions on Information Theory. Her doctoral research has been funded by several competitive fellowships, including the CyLab Security and Privacy Institute Presidential Fellowship, the College of Engineering Dowd Fellowship, and the Center for Informed Democracy & Social-cybersecurity Knight Fellowship. In addition to her passion for research and teaching, she strives to improve representation in computing. She led and organized the first Pittsburgh Women in Mathematics and Computing Symposium to provide mentorship and improve support structures for undergraduate students working at the interface of computing and mathematics. For her contributions to diversity, equity, and inclusion, she has been recognized with an Unsung Hero Award at CMU and the Advanced Graduate Ambassadorship of the Institute for Advanced Study (IAS), Princeton.